Available online at www.sciencedirect.com
Geochimica et Cosmochimica Acta 74 (2010) 4219–4233
www.elsevier.com/locate/gca
High- and low-affinity binding sites for Cd on the bacterial
cell walls of Bacillus subtilis and Shewanella oneidensis
Bhoopesh Mishra a,*, Maxim Boyanov c, Bruce A. Bunker a, Shelly D. Kelly c,
Kenneth M. Kemner c, Jeremy B. Fein b
a
Department of Physics, University of Notre Dame, Notre Dame, IN 46556, USA
Department of Civil Engineering and Geological Sciences, University of Notre Dame, Notre Dame, IN 46556, USA
c
Molecular Environmental Science Group, Biosciences Division, Argonne National Laboratory, Argonne, IL 60439, USA
b
Received 31 August 2009; accepted in revised form 17 February 2010; available online 23 February 2010
Abstract
Bulk Cd adsorption isotherm experiments, thermodynamic equilibrium modeling, and Cd K edge EXAFS were used to
constrain the mechanisms of proton and Cd adsorption to bacterial cells of the commonly occurring Gram-positive and
Gram-negative bacteria, Bacillus subtilis and Shewanella oneidensis, respectively. Potentiometric titrations were used to characterize the functional group reactivity of the S. oneidensis cells, and we model the titration data using the same type of nonelectrostatic surface complexation approach as was applied to titrations of B. subtilis suspensions by Fein et al. (2005). Similar
to the results for B. subtilis, the S. oneidensis cells exhibit buffering behavior from approximately pH 3–9 that requires the
presence of four distinct sites, with pKa values of 3.3 ± 0.2, 4.8 ± 0.2, 6.7 ± 0.4, and 9.4 ± 0.5, and site concentrations of
8.9(±2.6) 105, 1.3(±0.2) 104, 5.9(±3.3) 105, and 1.1(±0.6) 104 moles/g bacteria (wet mass), respectively. The
bulk Cd isotherm adsorption data for both species, conducted at pH 5.9 as a function of Cd concentration at a fixed biomass
concentration, were best modeled by reactions with a Cd:site stoichiometry of 1:1. EXAFS data were collected for both bacterial species as a function of Cd concentration at pH 5.9 and 10 g/L bacteria. The EXAFS results show that the same types of
binding sites are responsible for Cd sorption to both bacterial species at all Cd loadings tested (1–200 ppm). Carboxyl sites are
responsible for the binding at intermediate Cd loadings. Phosphoryl ligands are more important than carboxyl ligands for Cd
binding at high Cd loadings. For the lowest Cd loadings studied here, a sulfhydryl site was found to dominate the bound Cd
budgets for both species, in addition to the carboxyl and phosphoryl sites that dominate the higher loadings. The EXAFS
results suggest that both Gram-positive and Gram-negative bacterial cell walls have a low concentration of very high-affinity
sulfhydryl sites which become masked by the more abundant carboxyl and phosphoryl sites at higher metal:bacteria ratios.
This study demonstrates that metal loading plays a vital role in determining the important metal-binding reactions that occur
on bacterial cell walls, and that high affinity, low-density sites can be revealed by spectroscopy of biomass samples. Such sites
may control the fate and transport of metals in realistic geologic settings, where metal concentrations are low.
Ó 2010 Elsevier Ltd. All rights reserved.
1. INTRODUCTION
Bacteria are ubiquitous in a wide range of low temperature aqueous systems, and aqueous metal adsorption onto
*
Corresponding author. Present address: Biosciences Division,
Argonne National Laboratory, Argonne, IL 60439, USA.
E-mail address: [email protected] (B. Mishra).
0016-7037/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.gca.2010.02.019
bacterial cell wall functional groups can influence metal
speciation, bioavailability, and transport (Ledin et al.,
1996, 1999; Ledin, 2000; Ohnuki et al., 2007). Metal
adsorption onto bacterial cell walls can be viewed as a surface complexation reaction, and the thermodynamic stabilities of a wide range of metal-bacterial surface complexes
have been determined using bulk adsorption measurements
(e.g., Fein et al., 1997; Small et al., 1999; Yee and Fein,
2001; Daughney et al., 2002; Yee and Fein, 2003; Borrok
4220
B. Mishra et al. / Geochimica et Cosmochimica Acta 74 (2010) 4219–4233
et al., 2004b; Wightman and Fein, 2005; Kulczycki et al.,
2005; Burnett et al., 2006; Leone et al., 2007; Ngwenya,
2007; Pokrovsky et al., 2008a). In addition, X-ray absorption spectroscopy studies have identified cell wall carboxyl,
phosphoryl, and sulfhydryl groups as important sites of metal binding (Hennig et al., 2001; Kelly et al., 2002; Panak
et al., 2002; Boyanov et al., 2003; Toner et al., 2005; Guiné
et al., 2006; Mishra et al., 2007; Pokrovsky et al., 2008b;
Mishra et al., 2009). Most of these previous studies examined metal–bacterial adsorption under conditions with relatively high metal:bacterial site concentration ratios, and
many of these studies focused on the pH dependence of
adsorption rather than on the concentration dependence.
While measurements of the pH dependence of metal
adsorption onto bacteria place constraints on the cell wall
sites responsible for the binding, measurements conducted
as a function of metal:ligand ratio better constrain the metal:site stoichiometry of the adsorption reactions than do
standard pH edge adsorption experiments. Furthermore,
the important binding mechanisms can change as a function of metal:ligand concentration ratio, and studies at relatively high and fixed metal concentrations may mask the
presence and importance of low concentration, high affinity
metal binding sites on surfaces (Sarret et al., 1998).
Extended X-ray absorption fine structure (EXAFS) spectroscopy has demonstrated that only a limited number of
binding sites are responsible for metal binding onto cell walls
from a range of bacterial species. Phosphoryl and carboxyl
functional groups are responsible for metal complexation
by both Gram-positive bacteria (Kelly et al., 2002; Boyanov
et al., 2003) and Gram-negative bacteria (Toner et al., 2005;
Hennig et al., 2001; Panak et al., 2002) despite fundamental
differences in molecular structure of their exterior surfaces.
These two functional groups have also been found responsible for metal binding in both natural and contaminated consortia of bacteria obtained from river water and a
manufactured gas plant site, respectively (Mishra et al.,
2009). However a Cd-sulfhydryl binding site was also identified in the river water consortium (Mishra et al., 2009). Using
infrared spectroscopy, Wei et al. (2004) studied both Grampositive and Gram-negative bacterial surface functional
groups and determined that phosphoryl and carboxyl groups
are the primary contributors to the negative charge, and
hence the reactivity of the bacterial cell wall.
Relatively few studies have investigated the mechanisms
of metal complexation with bacterial cell walls as a function
of metal loading. Sarret et al. (1998) examined Zn and Pb
sorption to fungal cell walls of Penicillium chrysogenum at
pH 6 as a function of Zn and Pb loading, and the phosphoryl group was found to be the predominant complexing
ligand. Carboxyl binding was significant only under the
highest Zn loading and the lowest Pb loading conditions.
Burnett et al. (2006) studied Cd adsorption onto the thermophilic species A. flavithermus, and found that at high
bacteria:Cd ratios Cd adsorption occurs by formation of
a 1:1 complex with deprotonated cell wall carboxyl functional groups. At lower bacteria:Cd ratios, a second
adsorption mechanism occurs at pH > 7, which may correspond to the formation of a Cd-phosphoryl, CdOH-carboxyl, or CdOH-phosphoryl surface complex. Guiné et al.
(2006) reported the importance of sulfhydryl ligands in
addition to phosphoryl and carboxyl ligands in the adsorption of Zn onto three Gram-negative bacterial strains
(Cupriavidus metallidurans CH34, Pseudomonas putida
ATCC12633, and Escherichia coli K12DH5a) at low loadings of Zn. Mishra et al. (2007) reported that sulfhydryl ligands, in addition to carboxyl and phosphoryl sites, were
required for EXAFS modeling of Cd binding with the
Gram-negative bacteria Shewanella oneidensis and an aquatic consortium of bacteria at 30 ppm Cd loading and 10 g/L
(wet mass) biomass. However, sulfhydryl ligands were not
required for EXAFS modeling of the Gram-positive bacteria Bacillus subtilis under the same conditions, consistent
with previous finding by Boyanov et al. (2003). Similarly,
Pokrovsky et al. (2008b) observed Cd complexation with
sulfhydryl functional groups on the marine anoxygenic bacteria R. palustris in systems with 3 ppm Cd and 4 g/L biomass. While the Cd-sulfhydryl binding by Mishra et al.
(2007) and Pokrovsky et al. (2008b) was observed in systems near pH 6, Mishra et al. (2009) demonstrated that
aquatic consortia of bacteria exhibit sulfhydryl binding of
Cd across a wide pH range and even at relatively elevated
Cd concentrations.
The objective of this study is to explore metal–bacterial
adsorption reactions using both bulk adsorption measurements and X-ray absorption spectroscopy over a wide
range of metal:ligand ratios, and to compare these results
for the Gram-positive bacterial species B. subtilis and the
Gram-negative bacterial species S. oneidensis. Cell wall
functional group reactivity for B. subtilis has been characterized by Fein et al. (2005), and we collect new potentiometric titration data in this study using S. oneidensis in
order to determine the site concentrations and acidity constants for the important binding sites on the S. oneidensis
cell wall. Thermodynamic equilibrium modeling of the bulk
adsorption data was used to constrain the stoichiometries
of the important proton and Cd adsorption reactions. EXAFS measurements were conducted at pH 5.9, over a wide
range of Cd concentrations (1–200 ppm for B. subtilis and
3–200 ppm for S. oneidensis) at a fixed bacterial concentration of 10 g/L (wet mass), and we use these data to constrain the molecular-scale Cd adsorption mechanisms and
to determine whether the adsorption mechanisms change
as a function of Cd loading onto the bacterial cell walls.
Although Cd binding onto B. subtilis cells has already
been studied using EXAFS by members of our group previously (Boyanov et al., 2003), the goals of this study are
fundamentally different. While Boyanov et al. (2003) studied the adsorption behavior of B. subtilis over a pH range
of 3.4–7.8 at a fixed Cd:biomass ratio, this study attempts
to determine the site stoichiometry, and the sequence of
reactions occurring on bacterial cell walls as a function of
Cd loading at a fixed pH.
2. METHODS AND MATERIALS
2.1. Bacterial growth and harvest
S. oneidensis str., formerly classified as S. putrefaciens
MR-1 (Venkateswaren et al., 1999), is a Gram-negative
High- and Low-affinity binding sites for Cd on bacterial cell walls
dissimilatory metal-reducing bacterium found in soils, sediments, surface waters, and ground waters (e.g., Sorenson,
1982; Nealson and Myers, 1992). B. subtilis is an aerobic
Gram-positive soil microorganism with well-characterized
surface charge and reactivity (Harden and Harris, 1952;
Beveridge and Murray, 1980; Fein et al., 1997,2005). The
bacterial growth and wash procedures used in this study
were identical for both bacterial species and similar to those
described in Fein et al. (2005). A plated colony was transferred from an agar plate to a test tube containing 7 mL
of sterilized trypticase soy broth (TSB) + 0.5% yeast extract. Each broth tube was then incubated for 24 h at
32 °C in an incubator/shaker. The subsequent bacterial suspension was then transferred to 1 L of TSB + 0.5% yeast
extract and incubated for another 24 h at 32 °C in the incubator/shaker. The cells were removed from the nutrient
medium by centrifugation and rinsed five times with
0.1 M NaClO4 (the electrolyte used in the experiments).
The cells were not acid washed in order to avoid disruption
of the cell wall structure (Borrok et al., 2004a). Biomass is
cited in terms of wet mass, determined by centrifugation at
5800g for 60 min, and corresponds to approximately eight
times the dry mass of the cells (Borrok et al., 2005).
2.2. Potentiometric titrations
Three replicate potentiometric titrations were performed, each using 100 g/L (wet mass) suspensions of
S. oneidensis in 0.1 M NaClO4. The electrolyte solution
was bubbled with N2 gas for 1 h prior to use, and the titration cell containing the bacterial suspension was kept under
a N2 atmosphere during the experiment. The sample was
continuously stirred with a small magnetic stir bar during
the titration. The titrations were first run down-pH from
an initial pH of approximately 4 with aliquots of 1.0005
N HCl to approximately pH 2.7, then up-pH with aliquots
of 1.005 N NaOH to approximately pH 9.5. Each addition
of acid or base occurred only after a stability of 0.01 °mV/s
was attained. Although it is likely that the buffering capacity of the bacterial cell wall extended to extremely low pH
conditions (e.g., Fein et al., 2005), we limited the potentiometric titrations to a low pH value of 2.7 in order to avoid
excessive disruption of the Gram-negative cell wall structure (Borrok et al., 2004a).
2.3. Cd adsorption experiments
Washed bacteria were suspended in high density polyethylene test tubes containing 0.1 M NaClO4 electrolyte
to form a suspension of 10 g/L of bacteria (wet mass). A
500 ppm parent solution of Cd in 0.1 M NaClO4 was prepared from a commercially-supplied 1000 ppm Cd reference
solution. The pH of this parent solution was adjusted to 5.9
by adding small aliquots of 1 M NaOH. Appropriate
amounts of the parent solution were added to the bacterial
suspensions, and these suspensions were topped off with
additional 0.1 M NaClO4 to achieve the desired Cd and
bacterial concentrations. The pH of each system was adjusted using small aliquots of 1 M HNO3 or NaOH, and
the systems were allowed to react for 2 h on a shaker.
4221
The concentrations of total Cd in the experimental systems
ranged from 0.25 to 200 ppm. Previous studies involving
B. subtilis have demonstrated that equilibrium of Cd
adsorption reactions occurs in less than 1 h, and that the
adsorption reactions are fully reversible (e.g., Fowle and
Fein, 2000). pH was monitored every 30 min, and adjusted
as required using minute aliquots of 1 M HNO3 or NaOH.
The final pH was measured after two hours, and the solution was then centrifuged. The final pH of each experimental system was 5.9 ± 0.1, and the pH of the solutions never
varied by more than 0.3 during the course of the experiments. The bacterial pellet was retained for X-ray absorption spectroscopy analysis, and the resultant supernatant
was filtered (0.45 lm) and analyzed for dissolved Cd using
an inductively coupled plasma-optical emission spectroscopy technique with matrix-matched standards. The concentration of metal adsorbed to bacteria in each vessel
was calculated by subtracting the concentration of metal
that remained in solution from the original Cd concentration in the experiment.
2.4. EXAFS measurements and data reduction
EXAFS is a powerful structural probe that provides
information on the short-range coordination environment
of the atom under study (Stern, 1974). Fluorescence mode
Cd K edge (26,711 eV) EXAFS measurements were performed at MRCAT, sector 10-ID beamline of the Advanced Photon Source at Argonne National Laboratory
(Segre et al., 2000). The energy of the incident X-rays was
scanned by using a Si(1 1 1) reflection plane of a cryogenically-cooled double-crystal monochromator. The third harmonic of the undulator was utilized, with an undulator
tapering of about 3.5 keV to reduce the variation in the
incident intensity to less than 15% over the scanned energy
range. X-rays of higher harmonic energies were removed
using grazing incidence reflection from a Pt-coated mirror.
The incident ionization chamber was filled with 100% N2
gas. The transmitted and reference ion chambers were filled
with 100% Ar gas. The fluorescence detector in the Stern–
Heald geometry (Stern and Heald, 1983) was filled with
Kr gas, and a Pd filter of six absorption lengths was used
to reduce the background signal. The incident X-ray beam
profile was 1 mm square. Linearity tests (Kemner et al.,
1994) indicated less than 0.1% nonlinearity for a 50% decrease in incident X-ray intensity. Several energy scans were
collected from each sample. All scans were aligned by using
the simultaneously collected absorption spectrum of a Cd
foil, by setting the first inflection point of the spectrum at
26,711 eV.
The homogeneous bacterial pellet that formed at the
base of each experimental test tube was loaded into a slotted Plexiglas holder, covered with Kapton film, and
transported immediately to the beamline for EXAFS measurements. All the EXAFS measurements were performed
within 30 h of the adsorption experiment, and the samples
were refrigerated during the entire time between adsorption
experiment and EXAFS measurements. Quick scans (continuous-scanning mode of the monochromator), with signal
sampling every 0.5 eV and an integration time of 0.1 s per
4222
B. Mishra et al. / Geochimica et Cosmochimica Acta 74 (2010) 4219–4233
point were used. The advantage of using quick-scans is that
it reduces the radiation exposure during a single scan. Consecutive spectra were monitored for possible radiation-induced changes; no changes were observed. To further
reduce the possibility of radiation-induced changes or of
sample inhomogeneity, the X-ray beam was moved to a
fresh spot every 5 scans. A total of 30–150 consecutive
scans from each sample were collected and averaged,
depending on the Cd concentration in the sample.
Data were analyzed using the UWXAFS package (Stern
et al., 1995). Processing of the raw data, including alignment of datasets and background removal, were implemented by using ATHENA (Ravel and Newville, 2005),
IFEFFIT (Newville, 2001) and AUTOBK (Newville
et al., 1993). The input parameter to ATHENA that determines the maximum frequency of the background, Rbkg,
was set to 1.1 Å (Newville et al., 1993). The data range used
for Fourier transforming the EXAFS v(k) data was 2.3–
9.8 Å1 with a Hanning window function and a dk value
of 1.0 Å1 (Newville et al., 1993). Simultaneous fitting of
each dataset with multiple k-weighting (k1, k2, k3) of each
spectrum was performed using the Fourier transformed
v(R) spectra. The fitting range for all of the datasets was
1.2–3.4 Å. The simultaneous fitting approach reduces the
possibility of obtaining erroneous parameters due to correlations at any single k-weighting (Kelly et al., 2002; Mishra
et al., 2009).
2.5. EXAFS standards
Powder and aqueous Cd standards were used to determine the spectral features of Cd in carboxyl, phosphoryl
and sulfhydryl binding environments. A CdS powder standard was prepared from commercially available chemicals
(Sigma–Aldrich), after grinding and sieving (400 mesh).
The aqueous Cd standards included Cd-acetate, and Cdphosphate solutions. All aqueous Cd standards were prepared by dissolving Cd(NO3)2.4H2O in the appropriate
acid/electrolyte. The Cd-acetate and Cd-phosphate standards were prepared with a Cd:ligand ratio of 1:100 (acetate
or phosphate), by adding appropriate amounts of acetic
and phosphoric acids. The pH of the Cd-acetate solution
was adjusted using NaOH to 4.5, where the aqueous Cdacetate aqueous complexes dominate the Cd speciation in
solution. The pH of the Cd-phosphate solution was kept
at 3.0 to avoid precipitation of a cadmium phosphate solid
phase. Although the choice of standards is of paramount
importance when applying a linear combination fitting approach to X-ray absorption near edge structure (XANES)
data, EXAFS analysis is model independent. The main
use of experimental standards is to fine-tune the fitting
parameters for a given signal against a theoretical EXAFS
signal generated by simulating a known crystal structure (as
explained below). Therefore, the choice of aqueous versus
crystalline standards for EXAFS analyses is not critical,
and the use of a crystalline CdS powder as a standard for
the biomass Cd–S EXAFS signal, when all other standards
are aqueous, is acceptable.
EXAFS data analysis is based on refining theoretical
EXAFS spectra against the experimental data. Models are
constrained by use of crystalline model compounds with
well-characterized local structures. The crystallographic
information of the standard compounds were first transformed into a cluster of atoms by using the program
ATOMS (Ravel, 2001). Then FEFF8 (Ankudinov et al.,
1998) was used to carry out self-consistent quantum
mechanical calculations to simulate theoretical EXAFS
spectra based on the cluster of atoms thus obtained. Experimentally obtained EXAFS data on the standard compounds were fit to these theoretically generated EXAFS
spectra using the program FEFFIT (Newville et al.,
1995). We refined ab initio calculations on clusters of atoms
derived from known crystal structures (Caminiti, 1982;
Caminiti et al., 1984) against the EXAFS data from powdered CdS and aqueous Cd-acetate and Cd-phosphate
standards.
The value obtained for the EXAFS amplitude reduction
factor for all standards was S 20 = 1.00 ± 0.03, and this value
was used in modeling the spectra from the bacterial samples. Statistically significantly lower R factor and v2m values
were used as criteria for improvement in the fit to justify the
addition of an atomic shell to the model (Kelly et al., 2002).
3. RESULTS AND DISCUSSION
3.1. Potentiometric titrations
The three replicate titration curves are similar and are
depicted in Fig. 1 as the mass normalized net molality of
H+ added versus pH. The mass normalized net molality
of H+ added is equal to the concentration of acid added
to the system minus the concentration of base added, divided by the mass of the bacteria (in g/L; wet mass) used
in the titration. Similar to other bacterial titrations (e.g.,
Plette et al., 1995; Fein et al., 1997; Daughney et al.,
1998; Cox et al., 1999; Yee and Fein, 2003; Yee et al.,
2004; Haas et al., 2001; Sokolov et al., 2001; Martinez
et al., 2002; Ngwenya et al., 2003; Smith and Ferris, 2003;
Claessens et al., 2004; Fein et al., 2005), our potentiometric
titration data indicate that S. oneidensis exhibits significant
proton buffering over the entire pH range studied. This
finding indicates that the bacterial cell wall was not fully
protonated, even under the lowest pH conditions attained
(2.7), and it is possible that proton-active functional
groups are important contributors to the buffering at even
lower pH values. The potentiometric titration data obtained by Haas et al. (2001), Sokolov et al. (2001), and
Smith and Ferris (2003) for S. putrefaciens exhibit positive
values for the mass normalized net molality of H+ added to
solution below pH values of approximately 7, with negative
values at higher pH. Similar to those previous studies, the
mass normalized net molality of H+ added in this study is
positive below a pH of approximately 7.5–8.0, and exhibits
negative values at higher pH conditions. Although some
previous studies have done so, the sign of this parameter
should not be interpreted to indicate the sign of the surface
charge of the bacterial surface at a particular pH value.
Rather, the excess or deficit of this parameter relative to
its value for a bacteria-free blank at the same pH indicates
only that protons have been taken up or released by the
High- and Low-affinity binding sites for Cd on bacterial cell walls
4223
(Net Molality H+ Added)/g
4.0E-04
3.5E-04
3.0E-04
2.5E-04
2.0E-04
1.5E-04
1.0E-04
5.0E-05
0.0E+00
-5.0E-05
-1.0E-04
-1.5E-04
2
3
4
5
6
7
8
9
10
pH
Fig. 1. Three replicate titration curves of S. oneidensis (50 g/L, wet wt.) in 0.1 M NaClO4. Open diamonds, circles, and triangles represent
experimental data points for individual titrations. Solid curve represents the average model fit.
bacteria. Electrophoretic mobility measurements indicate
an overall negative charge associated with the S. oneidensis
cell wall (Claessens et al., 2004). Therefore, we model the
charging behavior of S. oneidensis with a series of organic
acid functional group sites. We use a surface complexation
approach to model the potentiometric titration and Cd
adsorption data in order to relate pH and sorbent:sorbate
ratio effects on adsorption to the speciation of the bacterial
surface. Our first step is to model the potentiometric titration data to determine site concentrations and stability constants for the important surface reactive sites. A generalized
deprotonation reaction for bacterial surface sites is represented as:
þ
R Ai H ¼ R A
i þH
ð1Þ
where R–Ai represents a bacterial surface functional group
type and i is an integer P1. A generalized mass action equation for the above reaction is:
KaðiÞ ¼
þ
½R A
i aH
0
½R Ai H ð2Þ
where Ka and a represent the equilibrium (acidity) constant
and activity of the subscripted reaction or species, respectively, and the brackets represent the concentration of surface sites in moles/L of solution. Each bacterial site R–Ai
represents a discrete site with its own site concentration
and acidity constant. FITEQL (Westall, 1982) was used
as the computational tool to determine the number of discrete sites necessary to account for the observed buffering
behavior and to solve for the site concentrations and acidity
constants for each type of site. We used a non-electrostatic
surface complexation model to describe proton and Cd
adsorption onto the bacterial cell wall functional groups
(e.g., Fein et al., 2005).
For each model tested, FITEQL calculates a variance
function, V(Y), that describes the goodness-of-fit of the
model to the experimental measurements. The best-fitting
model was determined to be the one that yields the lowest
V(Y) value to the experimental data. Each titration was
modeled separately. A 4-site model with four deprotonation
reactions associated with each site provides the best fit to
the experimental data of all three bacterial titrations, with
an average V(Y) = 0.52. Models utilizing less than 4-sites
yield significantly higher V(Y) values and exhibit a worse
visual fit to the experimental data. Models utilizing more
than 4-sites fail to converge, indicating that the experimental data do not support more than four discrete sites. The
average model fit to all three bacterial titrations is illustrated as a solid curve in Fig. 1. The best-fitting reaction
stoichiometries with corresponding averaged deprotonation
constants and averaged site concentrations for the S. oneidensis bacterial surface are given in Table 1.
Borrok et al. (2005) reviewed and compiled the available
potentiometric titration datasets for individual bacterial
species, bacterial consortia, and bacterial cell wall components, and derived an internally consistent thermodynamic
model, similar to the model used in this study, for all the
datasets. Borrok et al. (2005) found that, if one assumed
a 4-site model for each titration dataset with pKa values
fixed to 3.1, 4.7, 6.6, and 9.0, the calculated site concentrations for the individual datasets were remarkably similar,
yielding best-fit values of 1.13 104, 9.08 105,
5.32 105, and 6.63 105 moles/g bacteria (wet mass)
for the site concentrations for the lowest to the highest
pKa sites, respectively. The site concentrations and
deprotonation constants that we determined for S. oneidensis (Table 1) are within uncertainties of the averages obtained by Borrok et al. (2005).
3.2. Cd adsorption
The results of the bulk adsorption measurements for B.
subtilis and S. oneidensis are shown in Fig. 2a and b, respectively. The observed extents of Cd adsorption for the two
Table 1
Log K and site concentration values of S. oneidensis surface sites.
[Site]a
8.9
1.3
5.9
1.1
a
(±2.6) 105
(±0.2) 104
(±3.3) 105
(±0.6) 104
Reaction
Log K
R–A(1)H0 ¡ R–A(1) + H+
R–A(2)H0 ¡ R–A(2) + H+
R–A(3)H0 ¡ R–A(3) + H+
R–A(4)H0 ¡ R–A(4) + H+
3.3 ± 0.2
4.8 ± 0.2
6.7 ± 0.4
9.4 ± 0.5
Site concentrations in moles/g.
4224
B. Mishra et al. / Geochimica et Cosmochimica Acta 74 (2010) 4219–4233
Fig. 2. Experimental data and models for different Cd:Site stoichiometries for (a) B. subtilis, and (b) S. oneidensis. These experiments were
conducted at pH 5.9 ± 0.2, and the 10 g/L bacterial solution was prepared in a 0.1 M NaClO4 electrolyte. The low Cd concentration range of
the experimental data has been expanded for clarity and shown as insets in Figs. 2a and b.
bacterial species are similar. For each species, the concentration of adsorbed Cd increases with increasing total Cd
in the experimental systems, but the slope decreases at higher Cd concentrations. Under the lowest Cd concentrations
studied, Cd is predominantly adsorbed to the solid phase,
while at higher total Cd concentrations a higher proportion
remains in solution. Such adsorption behavior is typical for
a system with a limited number of binding sites, in which
sites are undersaturated at low metal concentrations and
become increasingly saturated with increasing metal
loading.
3.3. Thermodynamic modeling of the adsorption isotherms
Adsorption experiments that are conducted as a function of solute:sorbent concentration ratio at a fixed pH
do not uniquely define which sorbing site or sites of a multi-site sorbent are responsible for the uptake, but the results
do place rigorous constraints on the average stoichiometry
of the adsorption reactions. A generic Cd adsorption reaction can be represented as an interaction between the aqueous Cd cation (Cd+2) and deprotonated surface sites on the
bacterial cell wall to create a bacterial surface complex:
ð2xyÞþ
xðCd þ2 Þ þ yðR A1
i Þ () ðR Ai Þy ðCdÞx
ð3Þ
where x and y represent stoichiometric coefficients that
must be determined experimentally. We use the discrete 4site non-electrostatic model of Fein et al. (2005) to describe
the protonation state of the cell wall functional groups for
B. subtilis; and we use the similar model for S. oneidensis
from this study. The pKa values for the four sites on B. subtilis are 3.3, 4.8, 6.8, and 9.1, respectively, and the corresponding site concentrations are 8.1 105, 1.1 104,
4.4 105, and 7.4 105 moles of sites/gm wet mass bacteria, respectively (Fein et al., 2005). We refer to these sites
as Sites 1–4, respectively.
Because the adsorption experiments were conducted under fixed pH conditions, the data can be equally well modeled by ascribing Cd adsorption onto any of the four site
types on the bacterial cell wall, and more than one site
may be involved in Cd binding under the experimental conditions. However, because the experiments were conducted
at pH 5.9, we choose to model the Cd adsorption data for
each bacterial species as an interaction between Cd2+ and
the deprotonated form of Site 2 (the site with a pKa value
of 4.8 for both B. subtilis, and S. oneidensis). This exercise
is intended only to constrain the average site stoichiometry
of the important Cd–bacterial complexes at pH 5.9, and
does not provide a molecular-scale understanding of the
Cd binding environment. We attempted to model the Cd
adsorption data using Cd:Site stoichiometries of 1:1, 1:2
and 2:1, using the V(Y) fit parameter from FITEQL to distinguish the best-fitting stoichiometry. The value of V(Y)
depends on estimates for standard deviation in the experimental data, but in general V(Y) values between 0.1 and
20 represent reasonable fits, and models with lower V(Y)
values represent better fits of the model to the data (Westall, 1982). Even if the reaction mechanism changes as a
function of Cd loading, that is if different binding sites
are important under different Cd loading conditions, this
approach determines the average reaction stoichiometry
for each binding mechanism.
The best-fitting models for each reaction stoichiometry
are shown in Fig. 2a and b. The low Cd concentration
range of the experimental data has been expanded for clarity and shown as insets in Fig. 2a and b. The V(Y) values
associated with the best-fitting 1:1, 1:2, and 2:1 Cd:Site stoichiometry models for B. subtilis are 10.6, 320.1, and 174.9,
respectively, indicating that the 1:1 model yields the best
overall fit to the experimental data over the entire Cd concentration range. Fig. 2a illustrates that the 1:1 model fit to
the data, with a calculated best-fit log K value for reaction
(3) of 3.4 ± 0.2, provides a better fit than do the 1:2 or the
2:1 models when considering the entire Cd concentration
range. The inset figure shows that the data cannot distinguish between the 1:1 and the 1:2 models under low Cd concentration conditions, so under these conditions, as the Cd
concentration decreases and as the site:Cd molal ratio increases, it is possible that a surface complex with a 1:2
Cd:site stoichiometry becomes important. Similarly for S.
oneidensis, the V(Y) values associated with the best-fitting
1:1, 1:2, and 2:1 Cd:Site stoichiometry models over the entire Cd concentration range studied are 20.9, 238.3, and
351.5, respectively. Fig. 2b demonstrates that the 1:1 model
fit to the S. oneidensis data, also with a calculated best-fit
High- and Low-affinity binding sites for Cd on bacterial cell walls
log K value for reaction (3) of 3.4 ± 0.3, provides a good fit
to the experimental Cd adsorption data over the entire
range of Cd concentrations. As is the case for the B. subtilis
data in Fig. 2a, the data cannot distinguish between the 1:1
and the 1:2 models under low Cd concentration conditions,
so it is possible that under these conditions a surface complex with a 1:2 Cd:site stoichiometry becomes important. It
should be noted that Cd may be bound to more than one
site under the experimental conditions, so the best-fit log
K values reported here represent averages of the stability
constant values for all of the important Cd–bacterial surface complexes.
The modeling results for B. subtilis and S. oneidensis are
remarkably similar. The two bacterial species exhibit similar extents of adsorption over the entire range of Cd concentrations studied, and because of their similar proton
reactivities and site concentrations, the calculated log K values for reaction (3) are within error of each other. Although
these data do not constrain which site is involved in Cd
binding, the 1:1 model yields the best fit to the entire Cd
concentration range dataset regardless of which site we
use in the model. If more than one site is involved in the
Cd binding over this range of Cd loadings, the modeling
suggests that any mechanism involves a 1:1 metal:site binding ratio, except perhaps under the lowest Cd concentration
conditions studied here where a 1:2 complex is also
possible.
3.4. Analysis of XAFS data from standard compounds
The fitting results for the Cd standards are shown in
Table 2. The coordination environment of Cd in the acetate-bearing standard was modeled with O and C shells,
corresponding to a bound acetate group. Data were fit with
5.5 (±0.3) O atoms in the first shell and 3C (fixed to this value based on speciation calculation) atoms in the second
shell, consistent with the bidentate bonding mechanism observed by CaminitI et al. (1984). The Cd–C distance in the
Cd-acetate standard was found to be 2.70 (±0.02) Å. The
aqueous Cd-phosphate standard data were fit with 5.8
(±0.3) O atoms in the first shell, and 1.5 (±0.3) P atoms
in the second shell. The Cd–O bond length was found to
be 2.28 (±0.02) Å, the same distance as in the Cd-acetate
standard, and the Cd–P bond distance was found to be
3.41 (±0.03) Å. The Cd sulfide standard was fit with 4 S
Table 2
Structural parameters obtained from fits of the standard compounds spectra.
Standard
Path
N
R(Å)
r2 (103 Å2)
CdAc
Cd–O
Cd–C
5.5 ± 0.3
3.0b
2.28 ± 0.02
2.70 ± 0.02
10.9 ± 0.9
12.8 ± 4.0
CdPO4
Cd–O
Cd–P
5.8 ± 0.3
1.5 ± 0.3
2.28 ± 0.02
3.41 ± 0.03
10.5 ± 1.2
15.0 ± 3.0
2.53 ± 0.02
4.2 ± 0.01
9.0 ± 1.0
25.0 ± 4.0
CdS
a
b
Cd–S
Cd–Cd
a
4.0
12.0a
Fixed to this value based on crystallographic data.
Fixed to this value based on speciation calculations.
4225
atoms in the first shell at a distance of 2.53 (±0.02) Å.
The sulfide standard displays characteristic spectral features
that can be seen in Fig. 3a. The second and third oscillations of the Cd sulfide spectrum are out of phase relative
to all the other standards. The phase shift arises from Cd
bonding to sulfur in the first shell, as opposed to bonding
to oxygen in all of the other standards. This phase shift
(and change in frequency) in the chi data is manifested in
a shift in the first peak of the Fourier transformed data
for the Cd–S spectrum in Fig. 4a relative to the other standard spectra. In the Cd-acetate data, note the reduction in
amplitude of the first shell peak in Cd-acetate compared to
the Cd-phosphate spectrum (Fig. 4a). Detailed modeling reveals that the reduction in amplitude is due to the signal
from the carbon atom interfering destructively with that
from the oxygen atom. The subtle features in the second
shell of the Cd-phosphate and Cd-acetate spectra are not
seen clearly from the full view of the magnitude of the Fourier transformed data, because C and P are both light elements that do not backscatter the photoelectrons
strongly. The enlarged real part of the Fourier transformed
data, however, illustrates the relatively small changes in the
second shell of the Cd-acetate and Cd-phosphate spectra
(Fig. 4b). As can be seen from Fig. 4b, the Cd-phosphate
standard exhibits a feature (line shape) at about 2.7 Å,
indicative of the P atom in a bound phosphate group.
The position of the S peak overlaps the position of the C
feature at 2.2 Å in the Cd-acetate spectrum, but is shifted
to larger distances. From the above analysis it is clear that
the characteristic spectral features of a Cd-carboxyl and a
Cd-phosphoryl local environment are: (1) for Cd-carboxyl
binding, a smaller first shell peak amplitude exists, associated with an increase in the second shell peak at about
2.2 Å (Fig. 4b), and (2) for Cd-phosphoryl binding, a larger
first shell amplitude is evident, along with a feature at 2.7 Å.
The characteristic feature of Cd-sulfhydryl binding is a
shifted first shell peak to larger distances relative to a Cd–O
environment, and this shift is associated with an increase
in the peak to a position just over 2.2 Å (Fig. 4b).
The collection of high quality EXAFS data allowed
assignment of the subtle features in the EXAFS spectra to
the corresponding ligands, as well as the quantitative modeling described above. The structural parameters of the
Cd–O, Cd–S, Cd–C and Cd–P contributions (paths) used
in fitting the Cd sulfide, Cd-acetate and Cd-phosphate standards are the same as listed in Table 3 of Mishra et al., 2009.
The presence of these contributions in the spectrum of an
unknown sample can be taken as indicative of sulfhydryl-,
carboxyl-, and phosphoryl- binding environments, respectively. Our approach was to use these well-calibrated paths
in the analysis of the spectra for the bacterial samples. A
Cd–Cd contribution (path) obtained from the fit of the
CdS spectrum was used to test for the presence of Cd precipitation in each sample.
3.5. Qualitative analysis of spectra from the bacterial
samples
The magnitude and real part of the Fourier transform
data for the B. subtilis samples at various Cd loadings are
4226
B. Mishra et al. / Geochimica et Cosmochimica Acta 74 (2010) 4219–4233
the amplitude oscillation and phase shift towards higher r
value at about 2.2 Å occurs in the real part of the Fourier
transform data (Fig. 4f) with decreasing Cd loading on
the bacterial cell wall. By the same arguments as those applied to the B. subtilis spectra, we conclude that an increase
in the relative amount of Cd atoms bound to sulfhydryl
groups on the S. oneidensis cell wall occurs with decreasing
Cd loading.
3.6. Quantitative EXAFS modeling
Fig. 3. (a) k3 weighed v(k) data for (a) the Cd standards, (b) B.
subtilis isotherm, and (c) S. oneidensis isotherm data. Data range
used for Fourier transform was 2.3–9.8 k (Å1).
shown in Fig. 4c and Fig. 4d, respectively. The data can be
compared to the standards spectra that are depicted in
Fig. 4a and b. An increase in the amplitude of the oscillation at 2.2 Å with decreasing Cd loading in the system
can be clearly seen in Fig. 4d. Based on the analysis in
the previous paragraph, this increase can be attributed to
either increasing carboxyl or sulfhydryl binding to Cd.
The choice between these two possibilities is based on the
shift of the main peak position towards higher r values in
Fig. 4c. As discussed above, this shift is characteristic of
sulfhydryl binding, and suggests that in the two lowest concentration samples, 3.0 and 1.0 ppm, the adsorbed Cd is
predominantly bound to sulfhydryl groups.
The magnitude and the real part of the Fourier transform data for the S. oneidensis samples are shown in
Fig. 4e and Fig. 4f, respectively. A consistent increase in
The six spectra from the samples taken along the B. subtilis isotherm were first fit independently, using the Cd–O,
Cd–S, Cd–C and Cd–P paths that were used to model the
Cd standards (Table 2). The numerical results from these
fits are not shown because the final model, based on simultaneous fitting of data from all six Cd concentrations,
yielded identical results but with smaller uncertainties. Below we discuss the individual fits separately only to show
that the observed binding behavior with Cd concentration
is not the result of the global fitting constraint.
We attempted to model the 200 ppm Cd B. subtilis spectrum with a single binding site and found that Cd-phosphoryl binding yields the best fit to the data. However,
the addition of a Cd-carboxyl binding site to the Cd-phosphoryl model significantly improves the fit. There is no evidence for Cd-sulfhydryl binding in the 200 ppm Cd
spectrum. Best-fit values of the path parameters (r2 and
DR) for this sample were the same (within uncertainties)
as those obtained for the Cd standards reported above.
Modeling of the 100 ppm Cd B. subtilis spectrum yields results similar to those from the 200 ppm Cd spectrum, with
phosphoryl and carboxyl binding able to yield an excellent
fit to the data and with no evidence for Cd-sulfhydryl binding. Conversely, modeling of the individual spectra for the
30, 15, 3 and 1 ppm Cd samples requires an additional Cdsulfhydryl binding site, and indicates an increase in the
sulfhydryl coordination numbers with decreasing Cd concentration. While the carboxyl contribution to Cd binding
remains constant (within error) over this concentration
range, the phosphoryl contribution remains constant (within error) up to 15 ppm, decreases for 3 ppm, and was not
required at all in order to successfully model the 1 ppm
data. A steep increase in the sulfhydryl coordination number was found at 1 ppm, indicating the predominance of
Cd-sulfhydryl binding under these conditions.
The Cd-sulfhydryl binding that we observed in the
30 ppm B. subtilis experiment, although only a small portion of the bound Cd budget, is in contrast with the findings
of Boyanov et al. (2003) and Mishra et al. (2007) who did
not report any sulfhydryl binding for the same bacterial
species at similar Cd concentrations and pH values. However, the S coordination number that we report here for
these conditions (0.08 ± 0.04) is extremely small, and could
easily be masked by carboxyl and phosphoryl functional
groups with small changes in experimental conditions.
Although independent modeling of the six B. subtilis
samples provides an approximation of the contributions
of Cd-carboxyl, -phosphoryl and -sulfhydryl binding in
each spectrum, the average binding environment is complex
High- and Low-affinity binding sites for Cd on bacterial cell walls
4227
Fig. 4. (a) Fourier Transform magnitude of the Cd Standard data, (b) real part of the Fourier transform data of the Cd standard data, (c)
Fourier Transform magnitude of the B. subtilis isotherm data, (d) real part of the Fourier Transform data of the B. subtilis isotherm, (e)
Fourier Transform magnitude of the S. oneidensis isotherm data, and (f) real part of the Fourier Transform data of the S. oneidensis isotherm.
in nature and there are overlapping contributions of relatively small amplitude in the spectra. Correlations within
each individual fit between coordination numbers and sigma-square values, as well as between DE0 (energy shift)
and bond distances, can mask trends in the structural
parameters in the series of samples. To deal with the problem, all six spectra were fit simultaneously, each at three
k-weights, resulting in a simultaneous fit of eighteen datasets. The energy shift, DEo, the radial distance (R) and
the distance variation (r2-value) to each ligand were refined,
but were constrained to be the same in all samples. This approach carries in it the implicit assumption of identical
molecular binding geometry to the corresponding ligands
in all samples, and attributes the variation in amplitude
to variations in the relative coordination numbers. In other
words, it is assumed that the Cd–ligand distance and variation in average radial distance (r2-values) are the same
whether Cd is bound to that ligand alone or is in a mixed
environment of Cd-carboxyl, Cd-phosphoryl and/or Cdsulfhydryl binding. This simultaneous fitting approach
yields best-fit values of the path parameters (r2 and DR)
for all paths that are the same (within uncertainties) as
those obtained for the Cd standards (Table 2). These variables were therefore fixed to be equal to the values obtained
for them from the Cd standards. The increased number of
independent data points for the fit and the reduced number
of variables resulted in smaller correlations between the fitting parameters, allowing their determination with smaller
uncertainties and revealing the trends in average binding
environment with changes in Cd loading.
Further confidence in the fitting approach adopted in
this study was gained by a simple test to determine the
robustness of the model. The variables (distances, and
Debye–Waller factors) which have been fixed in this fitting
approach to the values obtained by standard compounds
were set to the upper and lower limits of their uncertainties
and the fits were performed again (Table 2). This results in
less than a 15% change in the final coordination numbers of
the corresponding ligands reported in Table 3a. Although a
rigorous qualitative modeling of the data has been attempted, and a robust fitting of the EXAFS spectra has
been achieved, the determination of the absolute number
of ligands bound to a metal in a complex natural system
is difficult. However, the technique is much more precise
in determining relative changes in the coordination environment of the metal, so the observed changes in the coordination numbers that we report in Table 3 are significant and
real.
4228
B. Mishra et al. / Geochimica et Cosmochimica Acta 74 (2010) 4219–4233
Table 3
EXAFS fitting parameters for (a) B. subtilis and (b) S. oneidensis samples.
[Cd] (in ppm)
(a) Bacillus subtilis
Paths
No
Nc
Ns
Np
1.0
3.0
15
30
100
200
1.95 ± 0.16
1.22 ± 0.52
2.87 ± 0.14
3.93 ± 0.28
1.23 ± 0.62
0.98 ± 0.22
0.48 ± 0.24
4.67 ± 0.16
0.97 ± 0.48
0.17 ± 0.14
0.82 ± 0.36
4.78 ± 0.12
0.89 ± 0.46
0.08 ± 0.04
0.82 ± 0.32
4.98 ± 0.16
0.99 ± 0.60
4.93 ± 0.14
0.98 ± 0.48
1.12 ± 0.40
1.03 ± 0.40
3.18 ± 0.28
1.17 ± 0.42
1.42 ± 0.14
0.35 ± 0.26
3.72 ± 0.22
1.30 ± 0.56
0.90 ± 0.16
0.58 ± 0.30
4.28 ± 0.20
1.46 ± 0.60
0.45 ± 0.16
0.83 ± 0.30
4.98 ± 0.14
1.40 ± 0.54
2
2
Ro = 2.28 Å, ro = 0.009 Å
Rc = 2.70 Å, rc2 = 0.012 Å2
Rs = 2.53 Å, rs2 = 0.009 Å2
Rp = 3.41 Å, rp2 = 0.015 Å2
DE0 = 2.2 ± 0.8 eV
(b) Shewanella oneidensis
Paths
No
Nc
Ns
Np
2
4.10 ± 0.65
0.95 ± 0.34
2
Ro = 2.28 Å, ro = 0.009 Å
Rc = 2.70 Å, rc2 = 0.012 Å2
Rs = 2.53 Å, rs2 = 0.009 Å2
Rp = 3.41 Å, rp2 = 0.015 Å2
DE0 = 1.4 ± 0.6 eV
The number of variables used to fit the EXAFS data in
this study is less than one-third the number of independent
data points in the data as obtained using Nyquist’s theorem
(Nyquist, 1928). Hence, the goodness-of-fit of the model to
the data is not the result of excessive flexibility in the model
and each refined parameter represents meaningful information about the binding environments.
Final fitting parameters for the B. subtilis system are tabulated in Table 3a, and the magnitude and real part of the
Fourier transform of the data and fits are shown in Fig. 5a
and b, respectively. Table 3a shows that the number of oxygen atoms bound to each Cd atom consistently decreases
with decreasing Cd concentration from 4.93 ± 0.14 for the
200 ppm sample to 1.95 ± 0.16 for the 1.0 ppm sample.
This is concurrent with an increase in the average number
of sulfur atoms bound to each Cd atom from 0.8 ± 0.04
for 30 ppm to 2.87 ± 0.14 for the 1.0 ppm B. subtilis sample. The two highest Cd loading samples (200 and
100 ppm Cd) do not exhibit any sulfhydryl contribution
to the Cd binding. The coordination numbers of the carboxyl group (1.0 ± 0.5) remain the same within the uncertainty of the measurement over the entire concentration
range measured. The coordination number of the phosphoryl group remains the same (1.0 ± 0.3) within the
uncertainty of the measurement for the 200, 100, 30 and
15 ppm Cd samples, but decreases to 0.48 ± 0.24 for the
3.0 ppm sample. The 1.0 ppm B. subtilis data do not require
the inclusion of Cd-phosphoryl binding to account for the
experimental spectra.
We used a similar approach to model the S. oneidensis
data as we did for the B. subtilis data. That is, we first attempted to model each spectrum individually with as few
site types as possible. We then compared these results with
simultaneous fits of all of the spectra together. In general,
we observed similar Cd binding environments and trends
in the S. oneidensis samples as we did for the B. subtilis samples. Modeling of the 200 ppm Cd S. oneidensis spectrum
with a single binding site provided the best fit with Cdphosphoryl binding. However, as was the case for the
200 ppm Cd B. subtilis sample, the addition of Cd-carboxyl
binding to the Cd-phosphoryl model significantly improved
the fit. Best-fit values of the parameters (r2 and DR) for this
sample are the same (within uncertainties) as those obtained
for the Cd standards reported above. Modeling of the individual spectra for the 100, 30, and 15 ppm Cd samples
shows an increase in the sulfhydryl coordination numbers
with decreasing Cd concentration. While the carboxyl contribution to Cd binding remains constant (within error)
over this concentration range, the phosphoryl contribution
decreases with decreasing Cd concentration in the samples.
A steep increase in sulfhydryl coordination number to
4.10 ± 0.65 occurs for the 3 ppm Cd sample, where carboxyl and phosphoryl contributions to Cd binding are
not required to fit the data. This spectrum was fit solely
with a Cd–S path, indicating the complete dominance of
Cd-sulfhydryl binding under these conditions.
Final fitting parameters for the S. oneidensis system,
using the same simultaneous fitting approach as we applied
to the B. subtilis spectra, are listed in Table 3b, and the
magnitude and real part of the Fourier transform of the
measured spectra and the model fits are shown in Fig. 5c
and d, respectively. Table 3b shows that the number of oxygen atoms bound to each Cd atom consistently decreases
with decreasing Cd concentration from 4.98 ± 0.14 for the
200 ppm sample to 3.18 ± 0.28 for the 15 ppm sample. This
is concurrent with an increase in the average number of S
High- and Low-affinity binding sites for Cd on bacterial cell walls
4229
Fig. 5. Fourier Transform (a) magnitude and (b) real part of the EXAFS data and fits for the B. subtilis isotherm, and Fourier transform (c)
magnitude and (d) real part of the EXAFS data and fits for the S. oneidensis isotherm.
atoms bound to each Cd atom from 0.45 ± 0.16 for the
100 ppm Cd sample to 1.42 ± 0.14 for the 15.0 ppm Cd-S.
oneidensis sample. The highest Cd loading sample (200
Cd) does not exhibit any Cd-sulfhydryl contribution. The
coordination numbers of the carboxyl group remain the
same (1.30 ± 0.15) as a function of Cd loading within the
uncertainty of the measurement. The coordination number
of the phosphoryl group decreases from 0.95 ± 0.34 for the
200 ppm Cd sample to 0.35 ± 0.26 for the 15 ppm Cd-S.
oneidensis sample. The 3 ppm Cd-S. oneidensis data were
modeled with Cd-sulfhydryl binding only. Carboxyl and
phosphoryl contributions to the Cd binding were not required for modeling the 3 ppm Cd data.
3.7. Discussion of Cd adsorption on bacterial cell walls as a
function of Cd loading
The EXAFS fitting results are consistent with the modeling of the bulk adsorption data, and indicate that B. subtilis and the S. oneidensis cells exhibit broadly similar
binding mechanisms for Cd as a function of metal loading
under the conditions of this study. Phosphoryl and carboxyl binding are primarily responsible for Cd binding to
B. subtilis cells at higher and intermediate Cd loading conditions (200, 100, 30 and 15). Carboxyl and sulfhydryl binding become much more important at 3 ppm and sulfhydryl
sites become the most important binding sites at 1 ppm Cd
loading for B. subtilis. On the other hand, phosphoryl binding dominates Cd binding to S. oneidensis cells under high
Cd loading conditions (the 200 and 100 ppm Cd samples).
The importance of carboxyl functional groups is highest
at intermediate Cd loading conditions (the 30 and 15 ppm
Cd samples) for S. oneidensis, due to the decreasing importance of the phosphoryl group. Sulfhydryl sites completely
dominate the Cd binding for the S. oneidensis samples with
the lowest Cd loading (3 ppm Cd). Sulfhydryl binding likely
does not disappear with increasing Cd loading onto the
bacterial cell wall. Rather, carboxyl and phosphoryl binding dominate the bound Cd budget under these conditions,
swamping the signal of the relatively low abundance Cdsulfhydryl sites. The EXAFS fitting parameters indicate
that more than one site is involved simultaneously in Cd
binding under most of the conditions studied, and except
for the lowest Cd concentrations studied, the dominant
adsorption stoichiometry for each site type is a 1:1 Cd:Site
ratio. These results are consistent with the surface complexation modeling of the bulk Cd adsorption data, which also
indicates an average Cd:Site stoichiometry of 1:1 and 1:2 at
the high and low Cd concentration ranges of this study. The
increase in average coordination number under the lowest
Cd loading conditions is evident from the steepening of
the Cd adsorption isotherms (Fig. 2a and b) under these
conditions. In addition, if we determine the average adsorption stoichiometry that is consistent with only the data
points with the lowest three Cd concentrations for each
bacterial species, then we obtain best-fitting Cd:Site stoichiometries of 1:3 and 1:4 for the B. subtilis and the S. oneidensis data, respectively. These results are consistent with the
increase in S coordination number to 2.87 ± 0.14 and
4.10 ± 0.65 for the lowest Cd concentrations studied for
B. subtilis and S. oneidensis.
The lowest concentration S. oneidensis sample exhibits
sulfhydryl binding only, and therefore can be used to estimate the concentration of sulfhydryl sites on the bacterial
cell wall. Cd forms stable tetrahedral bonds with S, so
our observed coordination number of 4.10 ± 0.65 for S is
consistent with the absence of any carboxyl and/or phosphoryl binding of Cd in this sample. It would, therefore,
4230
B. Mishra et al. / Geochimica et Cosmochimica Acta 74 (2010) 4219–4233
be reasonable to assume that for the 3 ppm Cd S. oneidensis
sample, the concentration of sulfhydryl functional groups
on the cell wall can be approximated as four times the concentration of adsorbed Cd atoms. This approximation is
likely an underestimation, as not all sulfhydryl sites on
the cell wall are involved in Cd binding under these conditions. However, these are the first estimates of the sulfhydryl site concentration and are valuable as preliminary
constraints at least. Under these conditions, the concentration of adsorbed Cd is 2.2 106 moles of Cd per gram
(wet mass) of bacteria, yielding an estimate of the sulfhydryl site concentration of 8.8 106 moles of sites per gram
(wet mass) of bacteria. This site concentration represents
only approximately 2% of the total number of sites on the
cell wall, as determined by the potentiometric titrations.
The observation and characterization of the low abundance, high-affinity sulfhydryl sites in this study was made
possible by using very low metal-to-surface loading conditions. This avoided masking of the sulfhydryl spectral signal
by larger signals from the more abundant, lower-affinity sites
that become occupied under higher metal loading conditions.
A possible reason why purely sulfhydryl bonding has not been
observed in previous spectroscopy studies was the use of high
metal loadings in order to obtain good signal-to-noise ratio
during data collection. We were able to collect high quality
spectra from low concentration samples by utilizing a highflux undulator beamline and a Stern–Heald fluorescence
detector (Stern and Heald, 1983). The more intuitive approach of using a 13-element energy-dispersive Ge detector
provided inferior data quality above k 8 Å1. We explain
this by the fact that our system is composed predominantly
of low-Z elements, resulting in a low fluorescence background. Therefore, the advantage of removing the small background using energy-dispersive detectors is outweighed by the
limited count rate compared to ionization chambers. In addition, by optimizing the gasses, voltages, and the sensitivity
regimes of the incident and fluorescence ionization detectors
we were able to achieve a better linear response between the
Io and If detectors. This resulted in low-noise data from samples as dilute as 1 ppm Cd in the original solution phase.
reasonably explained by a 1:1 metal:site reaction stoichiometry over the range of metal loading conditions studied
here. The EXAFS results are consistent with the bulk
adsorption measurements, indicating monodentate binding
of Cd onto both the carboxyl and the phosphoryl sites. The
only conditions where multi-dentate binding was observed
in the EXAFS data were for the lowest Cd loading samples
for both species (1 ppm for B. subtilis and 3 ppm for S.
oneidensis), where the coordination number for the sulfhydryl binding site becomes significantly greater than one.
The EXAFS measurements, collected over two orders of
magnitude of Cd:site loading conditions, indicate similar
binding environments for the two bacterial species studied.
In addition, the EXAFS modeling indicates that three different sites are responsible for Cd binding to these two bacterial cell walls. Fig. 6 depicts the general predominance
regions for the three different types of sites as a function
of Cd loading for the S. oneidensis samples, and a similar
picture would apply to the B. subtilis samples as well. Phosphoryl and carboxyl binding play important roles at high
(100–200 ppm) and mid (15–30 ppm) Cd loadings, respectively, for the S. oneidensis samples. For the B. subtilis samples, phosphoryl and carboxyl sites exhibit overlapping and
equal contributions for the samples with Cd concentrations
of 15 ppm and higher. Carboxyl binding becomes more
important than phosphoryl binding at 3 ppm Cd loading
for B. subtilis samples. However, at the lowest Cd loadings
for both of the species studied here, sulfhydryl surface complexes dominate the adsorbed Cd budgets. Because the sulfhydryl binding sites out-compete other deprotonated sites
for Cd at low Cd concentrations, these sites must have a
higher binding affinity for Cd than the other sites that are
deprotonated at pH 5.9. However, the EXAFS data in this
study do not constrain the location or identity of the sulfhydryl groups that bind Cd onto the biomass. The sulfhydryl
sites may be a component of cell wall-bound metal regulator proteins, or they may represent a component of a toxicity response mechanism located on the cell walls, initiated
by the non-metabolizing cells with the small amount of
stored energy that remains after washing. Our EXAFS
4. CONCLUSIONS
This study highlights the strengths and limitations of both
bulk adsorption and EXAFS measurements of metal–bacterial complexation. Bulk metal adsorption measurements
conducted as a function of metal loading provide constraints
on the average adsorption reaction stoichiometry, and
enable quantification of the stability constants for the important metal-bacterial complexes. However, these measurements do not provide for direct identification of the
important site or sites involved in metal binding. Conversely,
EXAFS measurements can rigorously identify the important
adsorption sites, but are less precise for determining reaction
stoichiometry and stability constant values.
We use these complementary techniques to elucidate the
nature and extent of Cd adsorption onto typical Gram-positive and Gram-negative bacterial cell walls. The bulk Cd
adsorption measurements indicate that adsorption can be
Fig. 6. A schematic representation of the regions of predominance
of the three different Cd binding environments as a function of Cd
loading onto biomass as determined by EXAFS.
High- and Low-affinity binding sites for Cd on bacterial cell walls
analysis indicates that the sulfhydryl sites are limited in
abundance relative to other cell wall site types, their binding
stoichiometries with Cd are much higher than those found
for the other functional groups, and these multi-dentate
bonds bind Cd tightly. However, a precise determination
of the location and more general molecular structure of
the sulfhydryl sites require additional experimental and
analytical constraints.
Because of the overlapping ranges of the phosphoryl,
carboxyl, and sulfhydryl binding, we could not explicitly
determine the stability constant for each Cd–bacterial surface complex for each species studied here. However, because a single stability constant yielded an excellent fit to
the bulk Cd adsorption data over a wide range of Cd loading conditions, the value of this stability constant represents
the average of the individual stability constants of the Cdsulfhydryl, the Cd-carboxyl, and the Cd-phosphoryl bacterial surface complexes. Potentiometric titrations and Cd
adsorption measurements that explicitly probe conditions
under which these sites each dominate are needed in order
to explicitly determine the speciation and thermodynamic
stability of each complex.
The EXAFS results presented in this study demonstrate
the following: (1) there is a broad similarity between the Cd
binding environments of these two species over a wide range
of Cd concentrations at circumneutral pH conditions, and (2)
Cd-sulfhydryl binding dominates the adsorbed Cd budget
under low Cd loading conditions. The similarity in the extent
of bulk Cd adsorption of these two bacterial species over a
wide range of Cd loadings complements the recent finding
that diverse bacterial consortia exhibit broadly similar binding environments over a wide pH range at a fixed Cd loading
(Mishra et al., 2009; Johnson et al., 2007; Borrok et al.,
2004b). Metal-sulfhydryl binding likely occurs under the
higher Cd loading conditions studied here, but its presence
is masked to EXAFS probing by the relatively high-abundance of carboxyl and phosphoryl binding sites when metal:site molal concentration ratios are high. However, our data
indicate the important role of these sites in Cd binding to
both Gram-positive and Gram-negative bacterial cells under
the circumneutral pH conditions of many natural systems.
Since Cd represents typical trace metal binding (e.g., Fein,
2000), sulfhydryl binding is likely to control metal binding
under low loading conditions for a wide range of metals of
environmental and geologic interest. Our finding that sulfhydryl is the dominant binding ligand at lower Cd loadings suggests that these sites may also dominate metal binding by
bacteria in realistic geologic settings where metals are typically present in trace concentrations.
ACKNOWLEDGMENTS
Research funding was provided by a National Science Foundation through an Environmental Molecular Science Institute Grant
to the University of Notre Dame (EAR02-21966), and from a
Bayer Corporation Predoctoral Fellowship. S.D.K. and K.M.K.
were supported by the US Department of Energy office of Science
Biological and Environmental Research Division Environmental
Remediation Science Program. MRCAT is supported by the US
Department of Energy under contract DE-FG02-94-ER-45525
4231
and the member institutions. Use of the Advanced Photon Source
was supported by the US Department of Energy under contract W31-109-Eng-38. Some analyses were conducted using instruments in
the Center for Environmental Science and Technology at University of Notre Dame. Help in sample preparation from Jennifer
Szymanowski, and beamline setup help from Tomohiro Shibata
and Soma Chattopadhyay of MRCAT are greatly appreciated.
Two anonymous journal reviewers and Associate Editor Michael
Machesky made helpful suggestions and significantly improved
the clarity of this paper.
REFERENCES
Ankudinov A. L., Ravel B., Rehr J. J. and Conradson S. D. (1998)
Real-space multiple-scattering calculation and interpretation of
X-ray-absorption near-edge structure. Phys. Rev. B 58, 7565–
7576.
Beveridge T. J. and Murray R. G. E. (1980) Sites of metal
deposition in the cell wall of Bacillus subtilis. J. Bacteriol. 141,
876–887.
Borrok D., Fein J. B. and Kulpa C. F. (2004a) Proton and Cd
adsorption onto natural bacterial consortia: testing universal
adsorption behavior. Geochim. Cosmochim. Acta 68, 3231–
3238.
Borrok D. M., Fein J. B. and Kulpa C. F. (2004b) Cd and proton
adsorption onto bacterial consortia grown from industrial
wastes and contaminated geologic settings. Environ. Sci. Technol. 38, 5656–5664.
Borrok D., Fein J. B. and Turner B. F. (2005) A universal surface
complexation framework for modeling proton binding onto
bacterial surfaces in geologic settings. Am. J. Sci. 305, 826–853.
Boyanov M. I., Kelly S. D., Kemner K. M., Bunker B. A., Fein J.
B. and Fowle D. A. (2003) Adsorption of cadmium to Bacillus
subtilis bacterial cell walls: a pH-dependent X-ray absorption
fine structure spectroscopy study. Geochim. Cosmochim. Acta
67, 3299–3311.
Burnett P. G. G., Daughney C. J. and Peak D. (2006) Cd
adsorption onto Anoxybacillus flavithermus: Surface complexation modeling and spectroscopic investigations. Geochim.
Cosmochim. Acta 70, 5253–5269.
Claessens J., Behrends T. and Van Cappellen P. (2004) What do
acid–base titrations of live bacteria tell us? A preliminary
assessment. Aqua Sci 66, 19–26.
Caminiti R. (1982) Nickel and cadmium phosphates in aqueoussolution – cation anion complex-formation and phosphate–
H2O interactions. J. Chem. Phys. 77, 5682–5686.
Caminiti R., Cucca P., Monduzzi M., Saba G. and Crisponi G.
(1984) Divalent metal–acetate complexes in concentrated
aqueous-solutions – an X-ray-diffraction and NMR-spectroscopy study. J. Chem. Phys. 81, 543–551.
Cox J. S., Smith D. S., Warren L. A. and Ferris F. G. (1999)
Characterizing heterogeneous bacterial surface functional
groups using discrete affinity spectra for proton binding.
Environ. Sci. Technol. 33, 4515–4521.
Daughney C. J., Fein J. B. and Yee N. (1998) A comparison of the
thermodynamics of metal adsorption onto two common bacteria. Chem. Geol. 144, 161–176.
Daughney C. J., Sicilliano S. D., Rencz A. N., Lean D. and Fortin
D. (2002) Hg(II) adsorption by bacteria: a surface complexation model and its application to shallow acidic lakes and
wetlands in Kejimkujik National Park, Nova Scotia, Canada.
Environ. Sci. Technol. 36, 1546–1553.
Fein J. B. (2000) Quantifying the effects of bacteria on adsorption
reactions in water–rock systems. Chem. Geol. 169, 265–280.
4232
B. Mishra et al. / Geochimica et Cosmochimica Acta 74 (2010) 4219–4233
Fein J. B., Daughney C. J., Yee N. and Davis T. A. (1997) A
chemical equilibrium model for metal adsorption onto bacterial
surfaces. Geochim. Cosmochim. Acta 61, 3319–3328.
Fein J. B., Boily J. F., Yee N., Gorman-Lewis D. and Turner B. F.
(2005) Potentiometric titration of Bacillus subtilis cells to low
pH and a comparison of modeling approaches. Geochim.
Cosmochim. Acta 69, 1123–1132.
Fowle D. A. and Fein J. B. (2000) Experimental measurements of
the reversibility of metal–bacteria adsorption reactions. Chem.
Geol. 168, 27–36.
Guiné V., Spadini L., Sarret G., Muris M., Delolme C., Gaudet J.
P. and Martins J. M. F. (2006) Zinc sorption to three Gramnegative bacteria: combined titration, modeling, and EXAFS
study. Environ. Sci. Technol. 40, 1806–1813.
Harden V. and Harris J. O. (1952) The isoelectric point of bacterial
cells. J. Bacteriol. 65, 198–202.
Haas J. R., Dichristina T. J. and Wade, Jr., R. (2001) Thermodynamic of U(VI) sorption onto Shewanella putrefaciens. Chem.
Geol. 180, 33–54.
Hennig C., Panak P. J., Reich T., Rossberg A., Raff J., SelenskaPobell S., Matz W., Bucher J. J., Bernhard G. and Nitsche H.
(2001) EXAFS investigation of uranium(VI) complexes formed
at Bacillus cereus and Bacillus sphaericus surfaces. Radiochim.
Acta 89, 625–631.
Kelly S. D., Kemner K. M., Fein J. B., Fowle D. A.,
Boyanov M. I., Bunker B. A. and Yee N. (2002) X-ray
absorption fine structure determination of pH-dependent
U-bacterial cell wall interactions. Geochim. Cosmochim. Acta
66, 3855–3871.
Kemner K. M., Kropf A. J. and Bunker B. A. (1994) A lowtemperature total electron yield detector for X-ray absorption
fine structure spectra. Rev. Sci. Instrum. 65, 3667–3669.
Kulczycki E., Fowle D. A., Fortin D. and Ferris F. G. (2005)
Sorption of cadmium and lead by bacteria–ferrihydrite composites. Geomicrobiol. J. 22, 299–310.
Ledin M., Krantz-Rulcker C. and Allard B. (1996) Zn, Cd and Hg
accumulation by microorganisms, organic and inorganic soil
components in multi-compartment systems. Soil Biol. Biochem.
28, 791–799.
Ledin M., Krantz-Rulcker C. and Allard B. (1999) Microorganisms
as metal sorbents: comparison with other soil constituents in
multi-compartment systems. Soil Biol. Biochem. 31, 1639–1648.
Ledin M. (2000) Accumulation of metals by microorganismsprocesses and importance for soil systems. Earth-Sci. Rev. 51,
1–31.
Leone L., Ferri D., Manfredi C., Persson P., Shchukarev A.,
Sjoberg S. and Loring J. (2007) Modeling the acid–base
properties of bacterial surfaces: a combined spectroscopic and
potentiometric study of the Gram-positive bacterium Bacillus
subtilis. Environ. Sci. Technol. 41, 6465–6471.
Martinez R. E., Smith D. S., Kulczycki E. and Ferris F. G. (2002)
Determination of intrinsic bacterial surface acidity constants
using a donnan shell model and a continuous pK(a) distribution
method. J. Colloid Interface Sci. 253, 130–139.
Mishra B., Fein J. B., Boyanov M. I., Kelly S. D., Kemner K. M.
and Bunker B. A. (2007) Comparison of Cd binding mechanisms by Gram-positive, Gram-negative and consortia of
bacteria using XAFS. AIP Conf. Proc. 882, 343–345.
Mishra B., Boyanov M. I., Bunker B. A., Kelly S. D.,
Kemner K. M., Nerenberg R., Read-Daily B. L. and Fein
J. B. (2009) An X-ray absorption spectroscopy study of cd
binding onto bacterial consortia. Geochim. Cosmochim. Acta
73, 4311–4325.
Nealson K. H. and Myers C. R. (1992) Microbial reduction of
manganese and iron: new approaches to carbon cycling. Appl.
Environ. Microbiol. 58, 439–443.
Newville M. (2001) IFEFFIT: interactive XAFS analysis and
FEFF fitting. J. Sync. Rad. 8, 322–324.
Newville M., Ravel B., Haskel D., Rehr J. J., Stern E. A. and
Yacoby Y. (1995) Analysis of multiple-scattering Xafs data
using theoretical standards. Physica B 209, 154–156.
Newville M., Livins P., Yacoby Y., Rehr J. J. and Stern E. A.
(1993) Near-edge X-ray-absorption fine-structure of Pb – a
comparison of theory and experiment. Phys. Rev. B 47, 14126–
14131.
Ngwenya B. T., Sutherland I. W. and Kennedy L. (2003)
Comparison of the acid–base behaviour and metal adsorption
characteristics of a Gram-negative bacterium with other strains.
Appl. Geochem. 18, 527–538.
Ngwenya B. T. (2007) Enhanced adsorption of zinc is associated
with aging and lysis of bacterial cells in batch incubations.
Chemosphere 67, 1982–1992.
Nyquist H. (1928) Certain topics in telegraph transmission theory.
Trans. AIEE 47, 617–644.
Ohnuki T., Yoshida T., Ozaki T., Kozai N., sakamoto F.,
Nankawa T., Suzuki Y. and Francis A. J. (2007) Chemical
Speciation and association of plutonium with bacteria, kaolinite clay, and their mixture. Environ. Sci. Technol. 41, 3134–3139.
Panak P. J., Booth C. H., Caulder D. L., Bucher J. J., Shuh D. K.
and Nitsche H. (2002) X-ray absorption fine structure spectroscopy of plutonium complexes with Bacillus sphaericus.
Radiochim. Acta 90, 315–321.
Pokrovsky O. S., Martinez R. E., Golubev S. V., Kompantseva E.
I. and Shirokova L. S. (2008a) Adsorption of metals and
protons on Gloeocapsa sp. cyanobacteria: a surface speciation
approach. Appl. Geochem. 23, 2574–2588.
Pokrovsky O. S., Pokrovsky G. S. and Feurtet-Mazel A. (2008b) A
structural study of Cd interaction with aquatic microorganisms.
Environ. Sci. Technol. 42, 5527–5533.
Plette A. C. C., Vanriemsdijk W. H., Benedetti M. F. and
Vanderwal A. (1995) pH dependent charging behavior of
isolated cell-walls of a Gram-positive soil bacterium. J. Colloid
Interface Sci. 173, 354–363.
Ravel B. (2001) ATOMS: crystallography for the X-ray absorption
spectroscopist. J. Sync. Rad. 8, 314–316.
Ravel B. and Newville M. (2005) Athena, artemis, hephaestus: data
analysis for X-ray absorption spectroscopy using IFEFFIT. J.
Sync. Rad. 12, 537–541.
Sarret G., Manceau A., Spadini L., Roux J. C., Hazemann J. L.,
Soldo Y., Eybert-Berard L. and Menthonnex J. J. (1998)
Structural determination of Zn and Pb binding sites in
Penicillium chrysogenum cell walls by EXAFS spectroscopy.
Environ. Sci. Technol. 32, 1648–1655.
Segre C. U., Leyarovska N. E., Chapman L. D., Lavender W. M.,
Plag P. W., King A. S., Kropf A. J., Bunker B. A., Kemner K.
M., Dutta P., Duran R. S. and Kaduk J. (2000) The MRCAT
insertion device beamline at the advanced photon source. Sync.
Rad. Instrum: 11th US Conf. CP521, 419–422.
Small T. D., Warren L. A., Roden E. E. and Ferris F. G. (1999)
Sorption of strontium by bacteria, Fe(III) oxide, and bacteria–
Fe(III) oxide composites. Environ. Sci. Technol 33, 4465–4470.
Smith D. S. and Ferris F. G. (2003) Specific surface chemical
interactions between hydrous ferric oxide and iron-reducing
bacteria determined using pKa spectra. J. Colloid Interface Sci.
266, 60–67.
Sokolov I., Smith D. S., Henderson G. S., Gorby Y. A. and Ferris
F. G. (2001) Cell surface electrochemical heterogeneity of the
Fe(III)-reducing bacteria Shewanella putrefaciens. Environ. Sci.
Technol. 35, 341–347.
Sorenson J. (1982) Reduction of ferric iron in anaerobic, marine
sediment, and interaction with reduction of nitrate and
sulphate. Appl. Environ. Microbiol. 43, 319–324.
High- and Low-affinity binding sites for Cd on bacterial cell walls
Stern E. A. (1974) Theory of extended X-ray absorption fine
structure. Phys. Rev. B. 10, 3027–3037.
Stern E. A. and Heald S. M. (1983) Basic principles and
applications of EXAFS. Handbook Sync. Rad. 10, 995–
1014.
Stern E. A., Newville M., Ravel B., Yacoby Y. and Haskel D.
(1995) The Uwxafs analysis package – philosophy and details.
Physica B 209, 117–120.
Toner B., Manceau A., Marcus M. A., Millet D. B. and Sposito G.
(2005) Zinc sorption by a bacterial biofilm. Environ. Sci.
Technol. 39, 8288–8294.
Venkateswaren K., Moser D. P., Dollhopf M. E., Lies D. P.,
Saffarini D. A., MacGregor B. J., Ringelberg D. B., White D.
C., Nishijima M., Sano H., Burghardt J., Stackebrandt E. and
Nealson K. H. (1999) Polyphasic taxonomy of the genus
Shewanella and description of Shewanella oneidensis sp. nov.
Int. J. Sys. Bacteriol. 49, 705–724.
Wei J., Saxena A., Song B., Ward B. B., Beveridge T. J. and
Myneni S. C. B. (2004) Elucidation of functional groups on
4233
Gram-positive and Gram-negative bacterial surfaces using
infrared spectroscopy. Langmuir 20, 11433–11442.
Westall, J. C. (1982) FITEQL, a computer program for determination for chemical equilibrium constants from experimental
data. Version 2.0. Report 82-02 Dept. Chem., Oregon St. Univ.,
Corvallis, OR, USA.
Wightman P. G. and Fein J. B. (2005) Iron adsorption by Bacillus
subtilis bacterial cell walls. Chem. Geol. 216, 177–189.
Yee N. and Fein J. (2001) Cd adsorption onto bacterial surfaces: a
universal adsorption edge? Geochim. Cosmochim. Acta 65,
2037–2042.
Yee N. and Fein J. B. (2003) Quantifying metal adsorption onto
bacteria mixtures: a test and application of the surface
complexation model. Geomicrobiol. J. 20, 43–60.
Yee N., Fowle D. A. and Ferris F. G. (2004) A donnan potential
model for sorption onto Bacillus subtilis. Geochim. Cosmochim.
Acta 68, 3657–3664.
Associate editor: Michael L. Machesky